All workshops are held in Founders Hall Room 602. Snacks are provided.
Presentation slots are available for all three dates. Schar School students from all of our PhD programs are encouraged to present their research, including working papers and ongoing projects. Faculty members are invited to give feedback on your presentations.
These workshops take the format of “mini conferences” in which students present their ongoing research. Presenters receive constructive feedback from the faculty members and engage fellow students in a discussion of potential research topics.
2:05pm - 2:40pm
Joshua Lee, Public Policy: “Bias in Artificial Intelligence: What, How, and Why?”
Artificial intelligence, particularly the subset known as deep learning/neural networks, has shown itself to be just as fallible as human beings when it comes to bias in decision-making. At first blush, AI would appear like it would eliminate such bias. After all, while human beings might be biased against one another based on characteristics such as race, gender, religion, or a myriad of other reasons, the idea is that a cold, calculating computer algorithm should be able to make simple decisions without such bias. Unfortunately, this idea has proven itself dead wrong. Over the past two years, there has been an explosion of scholarly research in the emerging field of detecting bias in AI and its implications for the future. What kinds of bias exist? How do we detect it? Why does it even exist in the first place? With neural networks already making inroads into the fields of healthcare, transportation, defense, law, and even foreign policy, it's only a matter of time before every area of policy is impacted.
2:40pm - 3:00pm|
Comments and Discussion
3:00pm - 3:05pm
3:05pm - 3:35pm
Meng-Hao Li, Public Policy: “Applying Network Analysis to Science and Technology Program Evaluation and Development”
In this presentation, I will share four cases regarding how I apply network analysis to approach and analyze science and technology program evaluation and development issues. The first case is a scientific collaboration network (global network). I built the stochastic actor-based model to estimate how the CTSA intervention program affects collaboration formation. The second case is a knowledge sharing network (ego-centric network). I used the multilevel analysis (mixed effect models) to examine how characteristics of scientists (individual-level) and the CTSA program influence knowledge transfer between two scientists (tie-level). The third case is a program diffusion network (global network). I employed the random-effects parametric survival model to explore how healthcare provider referral networks are associated with the diffusion of the EHR incentive program. The fourth case is a text network (two-mode network). I applied link predictions and supervised machine learning methods to understand previously unknown connections between biomedical subjects in Zika and CRISPR research. The tools and packages that I used to preprocess and analyze the network data will also be shared in the presentation, e.g. R, Python, Stata, SQL and Amazon cloud computing.
3:35pm - 3:55pm
Comments and Discussion
Interested students may sign up through: https://docs.google.com/forms/d/e/1FAIpQLSc-BGG_ox6mmmj8sARli7fY3QEviKo5g-mpz4YV2NanmikJkQ/viewform?usp=pp_url
If you have any questions regarding the PhD Research Workshops, please contact Abu Bakkar Siddique (firstname.lastname@example.org) or Neslihan Kaptanoglu (email@example.com).Learn More